xref: /aosp_15_r20/external/ComputeLibrary/src/runtime/CL/functions/CLConv3D.cpp (revision c217d954acce2dbc11938adb493fc0abd69584f3)
1 /*
2  * Copyright (c) 2021 Arm Limited.
3  *
4  * SPDX-License-Identifier: MIT
5  *
6  * Permission is hereby granted, free of charge, to any person obtaining a copy
7  * of this software and associated documentation files (the "Software"), to
8  * deal in the Software without restriction, including without limitation the
9  * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or
10  * sell copies of the Software, and to permit persons to whom the Software is
11  * furnished to do so, subject to the following conditions:
12  *
13  * The above copyright notice and this permission notice shall be included in all
14  * copies or substantial portions of the Software.
15  *
16  * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
17  * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
18  * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
19  * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
20  * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
21  * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
22  * SOFTWARE.
23  */
24 #include "arm_compute/runtime/CL/functions/CLConv3D.h"
25 
26 #include "arm_compute/core/CL/ICLTensor.h"
27 #include "src/gpu/cl/operators/ClDirectConv3d.h"
28 
29 namespace arm_compute
30 {
31 using namespace arm_compute::experimental;
32 
33 struct CLConv3D::Impl
34 {
35     const ICLTensor                        *src{ nullptr };
36     const ICLTensor                        *weights{ nullptr };
37     const ICLTensor                        *biases{ nullptr };
38     ICLTensor                              *dst{ nullptr };
39     std::unique_ptr<opencl::ClDirectConv3d> op{ nullptr };
40 };
41 
CLConv3D()42 CLConv3D::CLConv3D()
43     : _impl(std::make_unique<Impl>())
44 {
45 }
46 
47 CLConv3D::~CLConv3D() = default;
48 
configure(const ICLTensor * src,const ICLTensor * weights,const ICLTensor * biases,ICLTensor * dst,const Conv3dInfo & conv3d_info)49 void CLConv3D::configure(const ICLTensor *src, const ICLTensor *weights, const ICLTensor *biases, ICLTensor *dst, const Conv3dInfo &conv3d_info)
50 {
51     configure(CLKernelLibrary::get().get_compile_context(), src, weights, biases, dst, conv3d_info);
52 }
53 
configure(const CLCompileContext & compile_context,const ICLTensor * src,const ICLTensor * weights,const ICLTensor * biases,ICLTensor * dst,const Conv3dInfo & conv3d_info)54 void CLConv3D::configure(const CLCompileContext &compile_context, const ICLTensor *src, const ICLTensor *weights, const ICLTensor *biases, ICLTensor *dst, const Conv3dInfo &conv3d_info)
55 {
56     ARM_COMPUTE_ERROR_ON_NULLPTR(src, weights, dst);
57     ARM_COMPUTE_ERROR_THROW_ON(CLConv3D::validate(src->info(), weights->info(), ((biases != nullptr) ? biases->info() : nullptr), dst->info(), conv3d_info));
58 
59     _impl->src     = src;
60     _impl->weights = weights;
61     _impl->biases  = biases;
62     _impl->dst     = dst;
63 
64     _impl->op = std::make_unique<opencl::ClDirectConv3d>();
65     _impl->op->configure(compile_context, _impl->src->info(), _impl->weights->info(), _impl->biases ? _impl->biases->info() : nullptr, _impl->dst->info(), conv3d_info);
66 }
67 
validate(const ITensorInfo * src,const ITensorInfo * weights,const ITensorInfo * biases,const ITensorInfo * dst,const Conv3dInfo & conv3d_info)68 Status CLConv3D::validate(const ITensorInfo *src, const ITensorInfo *weights, const ITensorInfo *biases, const ITensorInfo *dst, const Conv3dInfo &conv3d_info)
69 {
70     return opencl::ClDirectConv3d::validate(src, weights, biases, dst, conv3d_info);
71 }
72 
run()73 void CLConv3D::run()
74 {
75     ITensorPack pack;
76     pack.add_tensor(TensorType::ACL_SRC_0, _impl->src);
77     pack.add_tensor(TensorType::ACL_SRC_1, _impl->weights);
78     pack.add_tensor(TensorType::ACL_SRC_2, _impl->biases);
79     pack.add_tensor(TensorType::ACL_DST, _impl->dst);
80     _impl->op->run(pack);
81 }
82 } // namespace arm_compute
83